You are on page 1of 5

IJEIT Vol. 1 No.

1 Oct 2009

Performance Evaluation and Simulation of Mobile Ad-hoc Network Routing


Protocols

Vikas Singla ,Rakesh Singla ,Ajay Kumar


Abstract: A mobile Ad hoc network (MANET) is formed by a group of autonomous mobile nodes connected by
wireless links, in which there is no backbone infrastructure. Many routing protocols with competing features have
been developed for wireless ad hoc networks. These protocols have varying qualities for different wireless routing
aspects. Amongst the most popular ones are Ad hoc On-demand Distance Vector (AODV), Destination-Sequenced
Distance-Vector Routing protocol (DSDV), Dynamic Source Routing Protocol (DSR) , and Optimum Link State
Routing (OLSR). In this paper we present our observations regarding the performance comparison of the DSR,
DSDV and AODV protocols for CBR traffic in mobile ad hoc networks (MANETs). We perform extensive
simulations, using NS-2 simulator. In this paper, three main issues will be addressed. The first is ‘Which routing
protocol provides a better performance in Mobile Ad hoc Networks?’ This question addresses the overall
performance of each routing protocol . The second issue addresses the factors that influence the performance of
these routing protocols. Finally yet importantly, the major differences in the routing protocols under study will be
addressed. In trying to answer these questions, MANET scenarios with varying traffic loads and mobility scenarios
were modeled and the performance of AODV, DSR, and DSDV protocols with respect to packet delivery ratio, end-
to-end delay, and normalized routing load. The performance of the routing protocol will greatly depend on various
factors such as network load and mobility effect

Keywords--- MANET, Routing, Ad-hoc, Protocol, Performance, Simulation, AODV, DSR, DSDV.

I. INTRODUCTION the increasing popularity of mobile devices and wireless


networks over the past few years, wireless ad-hoc
Advances in wireless technology and portable computing
networks has now become one of the most vibrant and
along with demands for greater user mobility have
active fields of communication and networking research.
provided a major impetus toward development of an
emerging class of self-organizing, rapidly deployable
II. ROUTING PROTOCOLS
network architectures referred to as ad-hoc networks. An
ad-hoc network is a collection of wireless mobile nodes Routing in Mobile Ad-hoc Network has been a subject of
dynamically forming a temporary network without the aid extensive research over the past several years. Because of
of any established infrastructure or centralized the fact that it may be necessary to pass several hops
administration. Ad-hoc networks are expected to play (multi-hop) before a packet reaches the destination, a
important role in future commercial and military settings routing protocol is needed. Routing protocol has two
where mobile access to a wired network is either functions:
ineffective or impossible. Potential applications for this (i) Selection of routes for various source-destination
class of network include instant network infrastructure to pairs
support collaborative computing in temporary or mobile
environments, emergency rescue networks for disaster (ii) Delivery of messages to their correct
management, remote control of electrical appliance, destination.
communication systems such as IVC (Inter-Vehicle The second function is conceptually straightforward
Communications) [10], and mobile access to the global using a variety of protocols and data structures (routing
Internet. tables).
Ad hoc wireless networks are self-creating, self-
organizing, and self-administrating networks. Ad hoc Ad-hoc routing protocols can be classified based on
networks are used where wired network and mobile different criteria. Depending upon the routing mechanism
access is either unproductive or not feasible. A employed by a given protocol, it may fall under more than
fundamental problem in ad hoc networking is how to one class. Routing protocols for Ad-hoc networking can
deliver data packets among MNs efficiently without be classified into four categories viz. (i) Based on routing
predetermined topology or centralized control, which is information update routing mechanism (proactive or
the main objective of ad hoc routing protocols. Since table-driven, reactive or on-demand and hybrid
mobile ad hoc networks change their topology frequently, protocols), (ii) Based on the use of Temporal information
routing in such networks is a challenging task. Moreover, (Past Temporal and Future Temporal) for routing, (iii)
bandwidth, energy and physical security are limited. With Based on routing topology (Flat Topology, Hierarchical

49
IJEIT Vol. 1 No. 1 Oct 2009

Topology), (iv) Based on the Utilization of Specific other mobile computer of the collection. In addition, each
Resources (Power Aware Routing and Geographical mobile computer agrees to relay data packets to other
Information Assisted Routing) . computers upon request. This agreement places a
2.1 Dynamic Source Routing Protocol (DSR) [4, 6] premium on the ability to determine the shortest number
The Dynamic Source Routing (DSR) protocol is an on- of hops for a route to a destination we would like to avoid
demand routing protocol based on source routing. DSR unnecessarily disturbing mobile hosts if they are in sleep
protocol is composed by two “on-demand” mechanisms, mode. In this way a mobile computer may exchange data
which are requested only when two nodes want to with any other mobile computer in the group even if the
communicate with each other. Route Discovery and Route target of the data is not within range for direct
Maintenance are built to behave according to changes in communication.
the routes in use, adjusting them-selves when needed.
Along with those mechanisms, DSR allows multiple III. RANDOM WAYPOINT MOBILITY MODEL
routes to any destination, thus can lead easily to load The Random waypoint model is a random-based mobility
balancing or increase robustness .In the source routing model used in mobility management schemes for mobile
technique, a sender determines the exact sequence of communication systems. The mobility model is designed
nodes through which to propagate a packet. The list of to describe the movement pattern of mobile users, and
intermediate nodes for routing is explicitly contained in how their location, velocity and acceleration change over
the packet’s header. In DSR, every mobile node in the time. Mobility models are used for simulation purposes
network needs to maintain a route cache where it caches when new network protocols are evaluated. In random-
source routes that it has learned. When a host wants to based mobility simulation models, the mobile nodes move
send a packet to some other host, it first checks its route randomly and freely without restrictions. To be more
cache for a source route to the destination. In the case a specific, the destination, speed and direction are all
route is found, the sender uses this route to propagate the chosen randomly and independently of other nodes. This
packet. Otherwise the source node initiates the route kind of model has been used in many simulation studies.
discovery process. Two variants, the Random walk model and the Random
2.2 Adhoc On-Demand Distance Vector (AODV) [4, 5, 6, direction model are variants of the Random waypoint
7] model.
AODV is a purely reactive routing protocol. In this In this model, a mobile node moves from its current
protocol, each terminal does not need to keep a view of location to a randomly chosen new location within the
the whole network or a route to every other terminal. Nor simulation area, using a random speed uniformly
does it need to periodically exchange route information distributed between [vmin, vmax]. vmin refers to the
with the neighbor terminals. Furthermore, only when a minimum speed of the simulation, vmax to the maximum
mobile terminal has packets to send to a destination does speed. The Random Waypoint Mobility Model includes
it need to discover and maintain a route to that destination pause times when a new direction and speed is selected.
terminal. In AODV, each terminal contains a route table As soon as a mobile node arrives at the new destination, it
for a destination. A route table stores the following pauses for a selected time period (pause time) before
information: destination address and its sequence number, starting traveling again. A Mobile node begins by staying
active neighbors for the route, hop count to the in one location for a certain period of time (i.e. pause).
destination, and expiration time for the table. The Once this time expires, the mobile node chooses a random
expiration time is updated each time the route is used. If destination in the simulation area and a speed that is
this route has not been used for a specified period of time, uniformly distributed between [vmin, vmax]. The mobile
it is discarded. node then travels toward the newly chosen destination at
2.3 Destination Sequenced Distance-Vector Routing the selected speed. Upon arrival, the mobile node pauses
(DSDV) [8] for a specified period of time starting the process again.
The Destination Sequenced Distance Vector Protocol The random waypoint model is the most commonly used
(DSDV) is a proactive, distance vector protocol which mobility model in the simulation of ad hoc networks. It is
uses the Bellmann -Ford algorithm. DSDV is a hop-by- known that the spatial distribution of network nodes
hop distance vector routing protocol, wherein each node moving according to this model is non uniform. However,
maintains a routing table listing the “next hop” and a closed-form expression of this distribution and an in-
“number of hops” for each reachable destination. This depth investigation is still missing. This fact impairs the
protocol requires each mobile station to advertise, to each accuracy of the current simulation methodology of ad hoc
of its current neighbors, its own routing table (for networks and makes it impossible to relate simulation-
instance, by broadcasting its entries). The entries in this based performance results to corresponding analytical
list may change fairly dynamically over time, so the results. To overcome these problems, it is presented a
advertisement must be made often enough to ensure that detailed analytical study of the spatial node distribution
every mobile computer can almost always locate every generated by random waypoint mobility. The movement

50
IJEIT Vol. 1 No. 1 Oct 2009

trace of a mobile node using the Random Waypoint 4.3 Average End-To-End Delay
model is shown in figure 3.1. It is considered that a Average End-to-End delay (seconds) is the average time
generalization of the model in which the pause time of the it takes a data packet to reach the destination. This metric
mobile nodes is chosen arbitrarily in each waypoint and a is calculated by subtracting “time at which first packet
fraction of nodes may remain static for the entire was transmitted by source” from “time at which first data
simulation time. packet arrived to destination”. This includes all possible
delays caused by buffering during route discovery
latency, queuing at the interface queue, retransmission
delays at the MAC, propagation and transfer times. This
metric is significant in understanding the delay introduced
by path discovery.
V. SIMULATION AND PERFORMANCE ANALYSIS
In this paper, we have taken two different scenarios. In
the first scenario four different pause times have been
taken i.e. 0, 5, 10 and 15 seconds. In the second scenario,
no. of nodes have been varied and performance
comparisons has been made between AODV, DSR and
DSDV protocols. Identical mobility and traffic scenarios
are used across protocols to gather fair results.
Fig 3.1: Traveling pattern of a mobile node using Random 5.1 Test Scenario 1
Waypoint Mobility
In the first scenario, We have chosen the simulation of
IV. PERFORMANCE METRICS [2] 100 nodes in 1000X1000 square meter area. Parameters
Mobile ad hoc networks have several inherent of this scenario are summarized in table 5.1. We have
characteristics (e.g. dynamic topology, time-varying and used CBR sources that started at different times because
bandwidth constrained wireless channels, multi-hop we want to get a general view of how routing protocol
routing, and distributed control and management). Design behaves, rather than TCP sources which use flow control
and performance analysis of routing protocols used for and retransmission feature.
mobile ad hoc network (MANET) is currently an active Table 5.1: Simulation Parameters for Test Scenario 1
area of research. To judge the merit of a routing protocol, Parameter Value
one needs metrics—both qualitative and quantitative-- Number of nodes 100
with which to measure its suitability and performance. Maximum Speed 20 m/s
Specifically, this paper evaluates the performance Simulation time 200 sec
comparison of AODV, DSR and DSDV routing protocols Pause Time 0,5,10, 15 sec
on the following performance metrics: Average routing
Environment Size 1000x1000
overhead, Packet delivery ratio and end-to-end delay.
Packet Size 512 bytes
4.1 Average Routing Overhead
Average routing overhead is the total number of routing Traffic Type Constant Bit Rate
packets divided by total number of delivered data packets. Packet Rate 2 packet/sec (1 kbps)
This metric provides an indication of the extra bandwidth 5.2 Test Scenario 2
consumed by overhead to deliver data traffic. It is crucial In the second scenario, the traffic as well as the scenario
as the size of routing packets may vary. The routing file is generated in the similar manner. Parameters of this
overhead describes how many routing packets for route scenario are summarized in table 5.2. In third scenario,
discovery and route maintenance need to be sent in order no. of nodes is varied and all the other parameters have
to propagate the CBR packets. It is an important measure been constant.
for the scalability of a protocol. It for instance determines,
if a protocol will function in congested or low-bandwidth Table 5.2: Simulation Parameters for Test Scenario 2
situations, or how much node battery power it consumes Parameter Value
4.2 Packet Delivery Ratio
Number of nodes 25, 50, 75, 100
Packet delivery ratio is calculated by dividing the number
Maximum Speed 20 m/s
of packets received by the destination through the number
of packets originated by the application layer of the Simulation time 200 sec
source (i.e. Constant Bit Rate (CBR)). It specifies the Pause Time 0 sec
packet loss rate, which limits the maximum throughput of Environment Size 1000x1000
the network. The better the delivery ratio, the more Packet Size 512 bytes
complete and correct is the routing protocol. Traffic Type Constant Bit Rate

51
IJEIT Vol. 1 No. 1 Oct 2009

Packet Rate 2 packet/sec (1 kbps) idea of the extra bandwidth that is required to deliver the
data packets.
5.3 Impact on Packet Delivery Ratio Average routing ove rhe ad
Based on the scenarios mentioned above, different “Scene
files” as well as “Traffic files” were generated and 70000

comparisons were made. The performance graphs Fig 5.1 60000

Routing overhead
and 5.2 illustrates the Packet delivery ratio for AODV, 50000

DSR and DSDV protocols. 40000 AODV


DSR
It can be seen that: 30000 DSDV

 AODV has Packet delivery ratio independent of 20000

change in pause time or no. of nodes. AODV achieves a 10000

packet delivery ratio between 85 to 100% for both test 0


0 5 10 15
scenarios. Paus e tim e

Packet Delivery ratio


Fig 5.3 : Average routing overhead when varying Pause
100 time(Scenario 1)
# of packets received / #

90
80
of packets sent

70 Average routing ove rhead


60 AODV
50 DSR 35000

40 DSDV 30000

Routing overhead
30
25000
20
AODV
20000
10 DSR
15000
0 DSDV

0 5 10 15 10000

5000
Pause tim e
0
25 50 75 100

# of node s
Fig 5.1: Packet Delivery Ratio with varying Pause
time(Scenario 1)
Fig 5.4 Average routing overhead when varying no. of
Packet Delivery ratio nodes(Scenario 2)
It can be seen that:
120
 DSR exhibits the highest average routing overhead
# of packets received /#

100
because of its route cache property. It generates the
of packets sent

80
AODV
highest no. of routing packets but its loss of packets is
60 DSR also more. DSR’s routing load increases in both scenarios.
40
DSDV
 AODV’s routing overhead gradually increases in
20 case change in no. of nodes is applied. However, it starts
0 declining as pause time is increased.
25 50 75 100
 DSDV is independent of change in pause time and
# of nodes no. of nodes. Routing overhead is lowest and constant in
both test cases because of its table-driven nature.
Fig 5.2 Packet Delivery Ratio with varying # of However, it gradually increases a bit for change in no. of
nodes(Scenario 2) nodes.
 DSDV also has a stable packet delivery ratio for 5.5 Impact on End-To-End Delay
change in pause time and no. of nodes but it is the lowest Fig. 5.5 and 5.6 describes the average end-to-end delay of
among all three protocols. DSR, AODV and DSDV protocols. Average end-to-end
 Change in no. of nodes has no effect on DSR’s delay is the average time it takes a data packet to reach
Packet Delivery ratio. But for change in pause time, it from source to destination. As route failure occurs, it
declines rapidly but later gets stable. DSR also achieves a takes longer for a data packet to reach its desired
packet delivery ratio between 85 to 100% for change in destination resulting in higher end-to-end delay.
no. of nodes. When change in pause time , packet delivery
ratio falls between 45 to 80%.
5.4 Impact on Average Routing Overhead
Fig 5.3 and 5.4 Illustrates the performance graphs for
average routing overhead required by all three protocols
when subjected to varying scenarios. This metric gives an

52
IJEIT Vol. 1 No. 1 Oct 2009

DSDV performs better than AODV and DSR. AODV


Average end to end delay
follows DSDV closely for average routing overhead. For
4.5
average end-to-end delay metric, the performance of
4
DSDV and AODV is quite similar to each other. DSR
performs poorly for this metric. From the comparison and
end to end delay

3.5

3 analysis, it can be concluded that the overall performance


AODV
2.5
DSR
of AODV protocol is better among all three. DSR
2
DSDV performs averagely for change in pause time and no. of
1.5
nodes. DSDV performs worst for the variation of all test
1
scenarios.
0.5

0
Future work will be to evaluate the performance of
0 5 10 15 MANET routing protocols under TCP traffic under
pause tim e different Mobility models. New Protocols can also be
implemented to improve the performance.
Fig 5.5 Average end-to-end when varying Pause time
(Scenario1) REFERENCES
[1] Ahmed S. and Ramani A. K., “Exploring the
Average end to end delay Requirements for QoS in Mobile Ad hoc Networks,”
Journal of Information & Communication Technology
1.4
Vol. 1, No. 2, (Fall 2007) 01-09
1.2
[2] Aziz S. R. A., Endut N. A., Abdullah S. and Daud
end to end delay

1
M. N. M., “Performance evaluation of AODV, DSR and
AODV
0.8
DSR
DYMO routing protocol in MANET”, CSSR 08-09, 14 -
0.6
DSDV 15 March 2009.
0.4 [3] Broch J., Maltz D. A., Johnson D. B., Hu Y. C.,
0.2 and Jetcheva J., “A performance comparison of multi-hop
0 wireless ad hoc network routing protocols,” ,ACM/IEEE
25 50 75 100
International Conference on Mobile Computing and
# of nodes Networking (MOBICOM’98), October 1998, pp. 85–97.
[4] Das S. R., Perkins C. E., Royer E. M. and
Fig 5.6 Average end-to-end when varying no. of nodes Marina M. K., “Performance comparison of two on-
(Scenario 2) demand routing protocols for ad hoc networks,” IEEE
Performance graphs illustrate that: Personal Communications Magazine, special issue on
 End-to-end delay in AODV is not affected by change Mobile Ad Hoc Networks, vol. 8, no. 1, pp. 16–29,
in pause time. It is affected when no. of nodes is changed; February 2001
however, it gets stable as the no. of nodes is increased. Its [5] Huang R., Zhuang Y., Cao Q., “Simulation and
performance is at par with DSDV. Analysis of Protocols in Ad Hoc Network”, 2009
 DSDV performs better than DSR and AODV as far International Conference on Electronic Computer
as average end-to-end delay is concerned. End-to-end Technology © 2009 IEEE
delay in DSDV is independent of any change in pause [6] Perkins C. E., Ad Hoc Networking, ed. Addison-
time or no. of nodes. It is lowest and most stable in both Wesley, 2000
test scenarios. [7] Perkins C. E. and Royer E. M., “Ad-Hoc On-Demand
 DSR is the most affected. Its end-to-end delay is Distance Vector Routing, Mobile Computing Systems and
highest. Applications,” Proc. IEEE Workshop Mobile Computing
VI. CONCLUSIONS & FUTURE WORK Systems & Applications (WMCSA ’99), pp. 90-100, 1999.
This study was conducted to evaluate the performance [8] Perkins C. E. and Bhagwat P., “Highly Dynamic
three MANET protocols i.e. DSR, AODV and DSDV Destination-Sequenced Distance-Vector Routing (DSDV)
based on CBR traffic. These routing protocols were for Mobile Computers”, SIGCOMM 94 -8/94 London
compared in terms of Packet delivery ratio, Average England UK @ 1994 ACM 0-89791 -682-4/94/0008.
routing overhead and Average end-to-end delay when [9] Royer E. M. and Toh C. K., “A review of current
subjected to change in pause time and varying no. of routing protocols for ad hoc mobile wireless networks”
nodes. Simulation results show that AODV performs IEEE Personal Communications, 6(2):46–55, April 1999.
better in terms of packet delivery ratio. DSR has the [10] Singla Vikas, Singla Rakesh, “An Advance
second highest packet delivery ratio among the three Approach to Vehicular Ad-hoc Network Security”
protocols. DSDV performs the poorest in terms of packet International Conference on Sensors and Related
delivery ratio. In terms of average routing overhead Networks(SENNET-07)”, December 2007.

53

You might also like